Analysis of a site x year experiment with timothy polycross progeny

Summary Results from a cooperative breeding programme with timothy for the northern areas of Scandinavia are presented. The main aim of the programme was to identify genotypes for synthetic populations that are high yielding, adapted over the whole area and stable over years. A polycross comprising 12 genotypes from each of five sites within the region was formed. The subsequent 60 half-sib families and four reference varieties were then compared under sward conditions at the same sites. The trials lasted for three years and results for total dry matter yields are presented.Significant differences in yield between lines were found. There were also significant two and three factor interactions between lines, sites and years. A new two-step procedure is presented to select the genotypes to make up a synthetic variety. Firstly, a superiority measure (Q-value) is used to select a group of genotypes that are high yielding and stable over sites and years. Secondly, the GxE part of the Q-value is split into two terms. One measures adaptation to predictable environmental conditions. The other one measures stability to unpredictable changes in the environment. A plot of these two measures provides a tool for discarding unstable or poorly adapted genotypes. The statistical properties of the selection parameters are discussed.